search for: checks_2

...will emit the checks, clone the loop and add the noalias metadata. Optionally the class could also try to locate a prior versioning of the loop and amend the checks and the noalias metadata. (This is the idea of my original post to only version once and keep amending the checking block: checks_1+checks_2.) Adam > On May 24, 2015, at 9:35 PM, Nema, Ashutosh <Ashutosh.Nema at amd.com> wrote: > > It’s a good thought in general Adam, but I worried about following scenarios: > > 1) As Dibyendu already mentioned Check1 + Check2 is not very clear. If your intent is superset/union...

...optimization from *multiple* passes, neither pass would make the decision locally to version. Also, it’s probably beneficial to perform a single loop versioning even if multiple passes would like to check different accesses. E.g. rather than: Checks_1 / \ / \ OrigLoop Checks_2 \ / \ \ / \ \ NoAlias_1 NoAlias_2 \ | / \ | / \ | / Join But instead: Checks_1+Check_2 / \ / \ OrigLoop NoAlias_2 \ / \ / \ / Join This is effectively c...

...> > > > > Also, it’s probably beneficial to perform a single loop versioning > even if multiple passes would like to check different accesses. E.g. > rather than: > > > > > > Checks_1 > > > / \ > > > / \ > > > OrigLoop Checks_2 > > > \ / \ > > > \ / \ > > > \ NoAlias_1 NoAlias_2 > > > \ | / > > > \ | / > > > \ | / > > > Join > > > > > > But instead: > > > > > > Checks_1+Check_2 > > > / \ &g...

...> > > > > Also, it’s probably beneficial to perform a single loop versioning > even if multiple passes would like to check different accesses. E.g. > rather than: > > > > > > Checks_1 > > > / \ > > > / \ > > > OrigLoop Checks_2 > > > \ / \ > > > \ / \ > > > \ NoAlias_1 NoAlias_2 > > > \ | / > > > \ | / > > > \ | / > > > Join > > > > > > But instead: > > > > > > Checks_1+Check_2 > > > / \ &g...